Matching and Price Competition: Would Personalized Prices Help?

Georgy Artemov, a Ph.D. candidate, published a paper analyzing the possible welfare from the NRMP switching to a differentiated salaries model. It specifically focused on the effect of imperfect information of preferences.


Here's the abstract:
We analyze the generalized deferred-acceptance algorithm when
preferences are known with an error. This algorithm incorporates personalized
salaries and is considered as a replacement for the current
algorithm for National Resident Matching Program (NRMP). Maintaining
Bulow and Levin’s (2006) assumption on preferences, we show
that an error in preferences of a worker propagates through the algorithm,
leading to a change in the salary of every more productive
worker. Thus, relatively small individual errors accumulate toward the
top and may lead to highly distorted salaries for top workers the same
way as mild compression translates into highly compressed salaries on
the top in the Bulow and Levin study of the current NRMP algorithm.

The article is slightly mathy, it reads a lot like the textbook, but it's interesting nonetheless. He uses a lot of the assumptions from Levin's model we saw in class, and it goes into more depth about the deferred acceptance algorithm than we went over in class. It's an interesting perspective! Check it out at: http://www.econ.brown.edu/students/Georgy_Artemov/artemov_personalized_mistakes_IJGT_f.pdf